Coffee is one of Indonesia's leading commodities, with various types such as Arabica, Robusta, and Liberica, each possessing similar aromatic characteristics that make them difficult to distinguish manually. Therefore, a technology-based automatic system is needed to detect and classify coffee types based on aroma. This study aims to design and develop a prototype of an Electronic Nose (E-Nose) system using gas sensors and an Artificial Neural Network (ANN) with the Backpropagation algorithm. The system employs five gas sensors (MQ-7,
an ESP32 MQ-136, MQ-137, MQ-138, and TGS-822) connected to microcontroller, which transmits aroma data in real-time to Firebase. The data were collected from three types of coffee: Arabica Aceh Gayo, Robusta Bali, and Liberica Jambi. After training and testing, the model's performance was evaluated using accuracy, precision, recall, and F1-score metrics. The evaluation results showed that the model achieved an accuracy of 86.7%, with the best classification
performance for the Arabica Aceh Gayo class (precision of 90.9%, recall of 100%, and F1-score of 95.3%). Meanwhile, the Robusta Bali class achieved an Fl-score
of 82.4%, and Liberica Jambi achieved 81.8%. These findings indicate that the
ANN-based system can classify coffee types with a relatively high degree of
accuracy based on sensor data and has the potential to serve as an alternative
solution for automatic and objective coffee identification.
Keywords: E-Nose, Coffee Classification, Artificial Neural Network, Backpropagation, Gas Sensors, MSE, ESP32